Autophagy literally means "self-eating" at the subcellular level. In the same way that your body breaks down fat and protein reserves if starved for a long enough time, the cell does the same thing, delivering cytoplasm to the lysosome/vacuole for degradation and recycling of the resulting macromolecules. Autophagy occurs in all eukaryotes and the protein components of the autophagic machinery are conserved from yeast to mammals. The hallmark of this process is the formation of double-membrane cytosolic vesicles, autophagosomes that sequester cytoplasm. After completion, the autophagosomes fuse with the lysosome/vacuole to release the inner vesicle that is broken down, allowing access to the cargo. Autophagy is a starvation response, but it also plays a role in various developmental processes and is associated with a range of pathophysiological conditions: Defects in autophagy can lead to some types of cancer, heart disease and neurodegeneration including Alzheimer's, Parkinson's and Huntington's diseases. Autophagy is also involved in innate immunity, in eliminating invading pathogenic bacteria and viruses and in lifespan extension. Autophagy can be non-specific;however, specific types of autophagy play a role in various processes including the removal of damaged or superfluous organelles, and the elimination of certain pathogens. The cytoplasm to vacuole targeting pathway is another example of specific autophagy. There are many questions about autophagy that remain to be answered. For example, we want to determine how environmental signals are transduced into an autophagic response, what regulatory controls determine the switch between specific and non-specific types of autophagy, the mechanism of achieving cargo specificity, the origin of the sequestering vesicle membrane and the molecular details of vesicle formation. We are using yeast to investigate the molecular mechanism of autophagy;this is the best system for a molecular genetic and biochemical analysis of this complex process. Because of the high degree of conservation, however, the information we learn from yeast will be applicable to higher eukaryotes. At present, we have characterized over twenty autophagy-related (Atg) proteins, determining their location and stage of action in the autophagy process. We are now defining specific interaction domains, and assigning a temporal order of action to the proteins with the ultimate goal of understanding their biochemical function. We have also established an in vivo reconstitution system that will allow us to define the functions and characteristics of individual Atg proteins. This knowledge will provide important information about basic cell biological processes including the mechanisms that govern dynamic membrane rearrangements, and allow specific recognition of subcellular components essential aspects required to maintain discrete organelles. A full understanding of autophagy may also allow therapeutic modulation of a process that is implicated in a range of disease conditions.